Research map / Multi-agent systems and games / Cooperative multi-agent RL
Learning Fair Policies in Decentralized Cooperative Multi-Agent Reinforcement Learning
Matthieu Zimmer, Claire Glanois, Umer Siddique, Paul Weng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use the SOTO two-subnetwork architecture instead of Basic | PPO with GGF or α-fairness · Matthew Effect | Pareto efficiency and equity | Improved |
| Use SOTO with GGF instead of FEN | Cooperative MARL · Matthew Effect, CLDE | total income and minimum resources collected | Improved |
| Use SOTO with α-fairness instead of FEN | Cooperative MARL · Matthew Effect | minimum resources collected | Improved |
| Use SOTO with α-fairness instead of FEN | Cooperative MARL · Matthew Effect | coefficient of variation | Mixed |
| Use a centralized critic with GGF instead of SOTO with GGF | Cooperative MARL · Job Scheduling | coefficient of variation | Improved |
| Use WQMIX instead of other compared methods | Cooperative MARL · Job Scheduling | coefficient of variation | Hurt |
| Use cooperation instead of independent selfish policies | Traffic-light control · SUMO | global waiting time | Improved |
| Use SOTO with GGF instead of other compared methods | Traffic-light control · SUMO | global waiting time | Improved |
| Use fully decentralized SOTO with GGF instead of FEN or WQMIX | Traffic-light control · SUMO | coefficient of variation | Improved |
| Use SOTO with GGF instead of FEN or a random policy | Data-center control · Continuous actions | coefficient of variation | Improved |
| Remove the trained self-oriented policy by using a randomly initialized one (ablation) | SOTO · Matthew Effect | social welfare optimization | Hurt |
| Replace incremental switching with pretraining the self-oriented policy first (ablation) | SOTO · Matthew Effect, CLDE | data efficiency | Hurt |
Closest papers by meaning
- Fair Cooperation in Mixed-Motive Games via Conflict-Aware Gradient Adjustment (NeurIPS 2025 · similarity 0.88)
- FairDICE: Fairness-Driven Offline Multi-Objective Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- Cooperative Policy Agreement: Learning Diverse Policy for Offline MARL (AAAI 2025 · similarity 0.88)
- Controlling Behavioral Diversity in Multi-Agent Reinforcement Learning (ICML 2024 · similarity 0.88)
- Finite-Time Global Optimality Convergence in Deep Neural Actor-Critic Methods for Decentralized Multi-Agent Reinforcement Learning (ICML 2025 · similarity 0.88)
- HyperMARL: Adaptive Hypernetworks for Multi-Agent RL (NeurIPS 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
